You may have seen a cute little Twitter hack — popularised by Andy Baio — which allows you to roll back the years and recreate a decade-old Twitter timeline. The twist is that you’ll be seeing updates from people who you may not have been following at the time but discovered later.

Nostalgia aside, the same technique could be used by journalists to track what was being said by any particular group of interest at a particular point in time. Here’s how.

Step 1: Create a Twitter account just for your search

The nostalgia aspect of Baio’s link relies on the fact that you are seeing 10-year-old updates by all the people you follow now. The search is personalised for each user, based on the people that they follow.

To adapt this for a journalistic purpose we need to make sure it is limited to the accounts we want to filter on. One way to do that is to create a new Twitter account that only follows those accounts.

For example, if we wanted to know what a group of election candidates were tweeting about four or five years ago at a previous election, we might create a Twitter account just to follow those accounts and conduct these ‘archive’ searches.

If you have a Gmail account, you don’t even need to create a new email address – just put a period anywhere in your gmail address and responses will still go to the same inbox.

Once you’ve set up that account you can recreate that decade-old Twitter timeline for the accounts it is following. But what if you want to look at another period?

Step 2: Break down the URL and adapt the search to your purposes

Baio’s tweet links to a particular advanced Twitter search which combines a number of ingredients. Here’s the URL in full:

https://twitter.com/search?f=tweets&vertical=default&q=filter%3Afollows%20until%3A2008-05-25%20-filter%3Areplies&src=typd

First, we have the base URL for searching Twitter — https://twitter.com/search?

Everything that follows that is built from what are called key-value pairs, like this:

f=tweets

These pairs contain the ingredients of the search, including how you want the results to be displayed. And each of those pairs is separated by an ampersand.

It’s helpful to list each one separately:

• f=tweets
• vertical=default
• q=filter%3Afollows%20until%3A2008-05-25%20-filter%3Areplies
• src=typd

Three of these are easy to explain: the first specifies that we want to see tweets. You can change it to f=images if you want to see images, for example, or f=videos if you want to see videos. Other options include f=broadcasts, f=news, and f=users.

I’m not entirely certain about vertical=default, but I assume it means we want to see results ordered chronologically (the default). And src=typd simply means that our search was typed rather than, for example, generated by a hashtag,

So most of the work in this search is done in this:

q=filter%3Afollows%20until%3A2008-05-25%20-filter%3Areplies

The q= part indicates that this is the query. The rest is the search that you can see both on the blue banner at the top of the search results and in the search box in the upper right corner:

filter:follows until:2008-05-25 -filter:replies

(In the URL spaces are encoded as %20 and colons as %3A).

This search has three parts to it: firstly, it applies a filter to your search so that it only shows accounts you are following. Secondly, it limits results to those before May 25 2008. Thirdly, it excludes replies.

We can show replies as well, then, by deleting the last part:

filter:follows until:2008-05-25

And we can change the date range by changing the date. So for example if we wanted to look at tweets around the UK’s referendum on EU membership, on May 7 2015, we would change it to:

filter:follows until:2016-06-23

And we could narrow it further by adding a keyword like ‘EU’:

filter:follows until:2016-06-23 EU

With these two ingredients, then: a Twitter account which is only following those accounts we want to look back at, and an advanced search which specifies the latest date range we want covered, you can trace back what was being said by those accounts at any point in Twitter’s archive.